Evolution of Multi-Tier Transmission Towards 5G Li-Fi Networks
Ahmed F. Hussein1, Hany Elgala1 and Thomas D.C. Little21Electrical and Computer Engineering Department, University at Albany - State University of New York,
2Electrical and Computer Engineering Department, Boston University, Boston, MA, [email protected]
Abstract— A design framework is presented in thismanuscript for a novel visible light communications (VLC)-based multi-tier waveform. Conventionally, VLC waveforms aredesigned to target specific services. In services that requirehigh-speed access, multi-carrier modulation techniques, i.e.,orthogonal frequency division multiplexing (OFDM), is consid-ered. For lower-speed access services, single carrier modulationtechniques are considered such as phase-shift keying (PSK) orpulse-position modulation (PPM). The proposed design offersa universal- and receiver-independent multi-tier waveform thatis expected to serve the requirements of fifth-generation (5G)wireless networks and beyond, including high-speed connectiv-ity, sensing and positioning services. This allows a wide varietyof devices to extract a useful portion of the received waveformassociated to the targeted service while ensuring inter-service-interference-free operation. In addition, the proposed designaims for cooperative transmission and dimming control toenhance the lighting environment for better user experience.
The paper provides a detailed description of the designprocess and the experimental evaluation. The experimentalresults indicate that the designed waveform can offer dimmingcontrol over 60% of the light-emitting diode (LED) full dynamicrange, while maintaining bit-error rate (BER) of 7× 10−5 for64-quadrature amplitude modulation (64-QAM).
The rapid evolution of wireless technology including wire-less LANs and mobile telephony has led to unprecedentedspectrum requirements by wireless users. According to [1], amedian of 54% reported using the Internet and/or owning asmartphone across 21 developing countries. This percentagehas increased to 87% across 11 developed countries includ-ing the US. Although recent advances in fourth-generation(4G) Long Term Evolution (LTE) systems can possiblyenable higher data rates to end users, it is not considered as aviable solution for the fifth-generation (5G) era and beyond[2]. Visible light communications (VLC) is a promising newtechnology that can offer enhanced indoor coverage withhigh data rates needed for 5G mobile networks [3]. More-over, VLC is considered as an enabling technology for futureInternet-of-Things (IoT) deployments [4]. VLC as an indoortechnology is compatible with other proposed technologiesfor 5G such as mobile femtocells and spatial modulation[5]–[7]. Coexistence with radio frequency (RF) spectrumbands such as 2.4GHz, 5GHz and 60GHz is introduced as a
novel architecture based on hybrid 802.11 WLAN and Li-Finetworks to reduce spectrum crunch [3].
In light of this fast-growing connectivity demand, there is aneed for novel approaches that can simultaneously serve mul-tiple services in future VLC networks. Few works considermulti-service capability of VLC waveforms. For example,in reference [8] low-speed broadcasting VLC systems basedon simple pulsed waveforms such as on-off keying (OOK)are considered on top of high-speed access links based onorthogonal frequency division multiplexing (OFDM). How-ever, this work does not consider dimming control in theproposed design. Other research teams introduce the con-cept of transmitting sinusoidal beaconing waveforms fromindividual light-emitting diodes (LEDs) for indoor sensingand gesture control [9]. This work does not consider VLCfor high-speed access. Dimming and data communicationsare considered in RPO-OFDM [10] for gigabit VLC links.In this technique, OFDM samples are conditioned to forma pulse-width modulated (PWM)-like envelope through thereverse polarity (RP) concept, i.e., resembling PWM cyclesfor digital dimming control. Accordingly, the signal-to-noiseratio (SNR) is independent of the brightness over a wide dim-ming range. Another framework is designed and proposedin [11], in order to serve the potential of a universal multi-tier waveform that can support different types of service inaddition to dimming control and sensing.
We propose a novel multi-tier waveform supported by adesign framework and experimental evaluation. The designoffers a universal waveform that can provide, (1) beaconingsignals for location and indoor navigation, (2) communica-tion links with embedded systems in the illuminated areas,(3) high-speed Internet access, and (4) dimming control forbetter user experience.
The remainder of this paper is organized as follows. InSec. II, the waveform design and the frame structure arepresented. The transmitter and receiver design is explainedin Sec. III. Section IV provides details related to the exper-imental evaluation process and results. Section V concludesthe paper.
II. FRAME STRUCTURE
In this section, the multi-tier waveform design and theframe structure are presented. This novel design enables themulti-service capability of a universal waveform that can
Fig. 1: A PWM encoded beacon showing corresponding duty cyclesto beacon samples. In (a) to (d), beacon’s phase control from 0o
to 270o is shown to represent QPSK modulation as an example ofconvenient modulation techniques to low-end receivers.
serve different services. For instance, it offers two differentpaths for high and low capacity links to serve both high-speed streaming and IoT services. In addition, monitoringservices based on optical beacons are included. Furthermore,the frame enables dimming control capability which is con-sidered as an essential feature for energy conservation andextending the lifetime of LEDs.
-k/2
0
k/2
-k/2
0
k/2
-k/2
0
k/2
50% df
50% df
80% df
80% df
20% df
20% df
Beacon
Beacon
Beacon
BPM
(a)
(b)
Fig. 2: (a) A PWM encoded beacon placed within different framestructures with varying duty cycles to generate waveforms withvarying illumination levels. (b) BPM illustration with 8 differenttime-slots per frame, as an example, to modulate the transmittedbits based on PPM concept.
Optical beacons have been introduced for different po-sitioning services. Inspired by PWM digital-to-analog con-verter (PWM-DAC) as being the most popular DAC struc-ture, the designed beacons within the proposed frame struc-ture are PWM modulated. The duty cycles of a PWM signal
sample the desired analog signal. However, this modulatedPWM signal also carries PWM harmonics. The analog signalcan be extracted using a low-pass filter (LPF), a simplepassive RC-filter that removes the PWM harmonics and letsthrough the analog signal. For example, in Fig. 1 a duty cycleof 50% represents the beacon sample at the 0V level, whilean 85% duty cycle represents the sample at the maximumvoltage.PWM encoded beacons can be controlled to modulate lowcapacity bits for IoT communication.PWM cycles are rear-ranged based on the target phase to modulate bits streambased on phase-shift keying (PSK) modulation. In Fig. 1,the beacon is manipulated to represent different phases thatcorrespond to modulated bits based on quadrature phase-shift keying (QPSK) modulation, as an example. In otherwords, demodulating the beacons in Figs. 1a-1d results in abit pattern of “00”, “01”, “10” or “11”. In order to maintaindimming control capability, the PWM encoded beacons areallocated within a frame structure. The frame duration Tfis decided based on the required refresh rate by the servedpositioning service. The added PWM duty cycles after and/orbefore the beacon are controlled to result in the target framebrightness ratio (FBR). This is explained in more details inthe following section.
-k/2
0
k/2
Beacon Dim Dim
Fig. 3: ACO-OFDM samples are conditioned to formulate a PWM-like encoded beacon within a PWM-like dimming control frame.
Fig. 2 illustrates how another path of low capacity bits aremodulated based on the position of the beacon within the fullframe. This beacon-position modulation (BPM) is inspiredby the pulse-position modulation (PPM) concept. As shownin the figure, this is applicable to any target dimming value.
RPO-OFDM is one of the few works that considers dif-ferent dimming levels while maintaining reliable broadbandcommunication. In this technique, the OFDM samples areconditioned to form a PWM-like envelope, and thus theSNR is independent of the brightness over a wide dimmingrange. For these reasons, the design of the high capacity linkwithin the designed frame is inspired by RPO-OFDM. Thehigh capacity stream of bits can be modulated by any of theunipolar optical OFDM techniques such as asymmetrically-clipped optical (ACO)-OFDM or as introduced in literature,DC-biased optical (DCO)-OFDM can be used with differentsettings [12]–[14]. The OFDM samples are then processedto generate the target frame structure including the PWMencoded beacon and the dimming control PWM duty cycles.
Fig. 4: Proposed system transmitter illustrating waveform generation by MATLAB, as well as, the hardware blocks.
Fig. 5: Proposed system receivers for both low and high capacity bits embedded within the received waveform.
This is shown in Fig. 3 where the OFDM samples, shownas blue, comprise a PWM-like envelope for the beacon andthe dimming control duty cycles, shown as black.
III. WAVEFORM GENERATION AND DETECTION
In this section, details about generation and detection ofthe designed waveform are provided.
Fig. 4 shows the waveform generation in MATLAB. First,beacon signals are PWM modulated based on the identifica-tion numbers (IDs) of the LEDs. Then, high and low capacitybit streams are separated into two different paths to generateOFDM waveforms and modulate the phases of the beacons.
A PWM frame is generated with a fixed duty cycle basedon the target illumination level. The beacon is concatenatedwithin the dimming frame based on the BPM low capacitybits. The OFDM samples are then structured to generate aPWM-like envelope, resembling the required PWM frame.The signal is fed into a DAC which modulates a DC-biasedLED around a DC-operating point. The transmitted universalwaveform then can be captured by either a high-end receiveror a receiver with limited resources.
Fig. 5 shows a proposed receiver design for both thelow and high capacity bit streams. The same waveform iscaptured by both receivers. The high capacity link has a high-speed photo-detector (PD) while the low capacity link caneither use the same PD followed by a LPF circuit or use alow-speed PD that can do the filtering process. The filteringprocess is essential to capture the beacon signal frequency
while filtering out all other PWM and OFDM harmonics.Each captured signal goes through a transimpedance am-plifier (TIA) and an analog-to-digital converter for furtherprocessing using MATLAB. In software, the two paths areused to capture OFDM bits, IDs of the LEDs and lowcapacity bits.
Algorithm 1 shows details of the frame structure to betterdescribe the block diagram of Fig. 4. A set of parametersmust be defined including the PWM frequency fPWM, bea-con frequency fB and the target FBR. The beacons are thenPWM encoded by converting the beacon sample amplitudesto PWM-varying duty cycles. A phase shift is added basedon the presence of PSK modulated bits. The PWM fixedduty cycle df for the dimming control part of the frameis decided based on the target FBR. The beacon positionwithin the frame is controlled based on BPM informationbits and then the full frame is generated with OFDM samplessuperimposed over the PWM frame. The detection process ofAlgorithm 2 follows the block diagram explained previouslyin Fig. 5. Additional analysis of the spectral characteristicsand dimming capabilities of the waveform can be found inprior work [11].
IV. RESULTS AND DISCUSSIONAn experimental setup is constructed to evaluate the per-
formance of the designed waveform for the multi-tier feature.The setup (Fig. 6) illustrates the high and low capacitylinks based on the receiver used to capture the transmittedwaveform. A light source (Thorlabs 50mW LED635L) is
Input : OFDM bitsLow capacity bits
Output: Multi-tier waveform S(t)Initialization:Define beacon frequency fBDefine PWM frequency fpwmSet frame duration Tf // beacon refreshrate
Set beacon duration TB // Includes one ormore beacon cycles
Set target frame brightness ratio FBRDefine OFDM parameters // FFT lengthNFFT, Sampling Frequency fs
beacon duty cycles10 Generate PWM sampled beacon with varying duty
cycles (dB)11 df = 1
Tf−TB[Tf FBR − TB
2 ] // Caclulate
dimming responsible duty cycles12 Aggregate beacon and dimming frame13 if BPM Modulation then14 Locate beacon within frame based on BPM bits;15 else16 Locate beacon at the beginning of the frame;17 end18 Generate ACO-OFDM samples19 OFDM samples formulate a PWM-like frame and
modulated by the generated waveform through an arbitrarywaveform generator (AWG) (Rigol). The received waveformis captured by an amplified detector (PDA10A) to capturethe OFDM bits. The DC-out channel of a receiver module(New Focus 1601) is used to capture the low capacity bitsembedded within the transmitted waveform. The receivedwaveforms from both branches are captured using an os-cilloscope and then further processed using MATLAB. Theexperimental setup and waveform design specifications aresummarized in Table I. The trade-off between the OFDMsignal quality and dimming capability is captured in Fig.7. Here, the SNR of the generated OFDM samples are
varied between 12dB, 18dB and 24dB. A full PWM frameis generated with an embedded PWM encoded beacon and afixed duty cycle of 50%. The x-axis represents the receivedsignal power (Prx) of a PWM frame without OFDM sampleswhile the y-axis represents the signal power of a PWM-likeenvelope generated by OFDM samples.
ID and low capacity bits detection1 Capture received waveform from low-speed
receiver2 Apply received waveform to LPF, if needed3 Split into two paths // For ID and low
demodulator6 Aggregate low capacity bits7 Apply IFFT to get fB8 Detect LED ID9 Apply positioning algorithm // Using
multiple LEDsOFDM bits detection
10 Capture the received waveform from ahigh-speed receiver
11 Extract OFDM symbols from PWM frame12 OFDM demodulation13 Concatenate OFDM received bits
Algorithm 2: Received waveform detection algo-rithm.
TABLE I: System experimental specifications and waveform designparameters
Experimental setup specifications
AWGManufacturer / Model Rigol / DG5352
fs 1GSa/sBw 350MHz
Light sourceManufacturer / Model Thorlabs / LED635L
Wavelength 635nmTypical optical power 170mW
High-speed receiver
Manufacturer / Model Thorlabs / PDA10ABw 150MHz
Detector Si PINActive area 0.8mm2
Wavelength range 200-1100nmResponsivity 0.44A/W
Low-speed channel (DC)Manufacturer / Model New Focus / 1601
Measured Bw 150kHzWavelength range 320-1000nm
ScopeManufacturer / Model Rigol / MSO4054
Bw 500MHzMaximum sampling 4GSa/s
Power meter Manufacturer / Model Thorlabs / S121CWavelength range 400-1100nm
Design parametersTf 20msTB 2ms
fB(1,2) 1,2kHzfs 4MHz
NFFT 16fPWM 200kHz
(a)
(b)
Fig. 6: Experimental setup to illustrate the reception process usingtwo different receivers for high and low capacity bit streams in (a)and (b), respectively.
0.5 1 1.5 2 2.5
Prx
without OFDM (mW)
1
1.5
2
2.5
Prx
with O
FD
M (
mW
)
SNR=12 dB
SNR=18 dB
SNR=24 dB
(a)
12 18 24
SNR (dB)
0
20
40
60
80
100
Dim
min
g c
ontr
ol ra
nge (
%)
(b)
Fig. 7: Experimental impact of varying SNR of conditioned OFDMsamples on the dimming capability of the transmitted waveformframe in (a) and effect on dimming control range in (b).
The received power is measured at a fixed distance of50cm using Thorlabs PM100USB power meter and S121C
10 15 20 25
SNR (dB)
10-6
10-4
10-2
100
BE
R
64-QAM
32-QAM
16-QAM
Fig. 8: Experimental BER of reconstructed and demodulated OFDMsamples.
(a) (b)
Fig. 9: (a) Captured frame by the high-speed receiver illustratingthe PWM encoded beacon, OFDM samples and fixed 50% dutycycles for dimming control. (b) Captured 20kHz beacon signals bythe low-speed receiver after concatenation from different frames.
PD sensor. Fig. 7a shows a slight variation between thereceived power of the PWM frame without OFDM andthe PWM-like frame with OFDM at SNR of 12dB. Thisis reflected by in Fig. 7b which shows the reduction indimming range control due to the reduction of PWM-likepeak-to-peak voltage as a result of OFDM sample generation.As the SNR increases, the variation in the received powerincreases. The obtained dimming control range for SNR of12dB is nearly 93% where the received signal power of aPWM-like envelope with OFDM, i.e., y-axis, varies from0.51mW to 2.45mW compared to a range between 0.51mWto 2.6mW for a PWM frame without OFDM. Similarly,dimming control ranges of 87% and 60% are obtained forSNR values of 8dB and 24dB, respectively. Though, thereis a clear trade-off between the OFDM signal quality andthe dimming range. SNR and dimming range are inverselyproportional; dimming control swing of 60% of the dynamicrange is still achievable at a high SNR value of 24dB.Next, the high-speed link is evaluated based on calculatingthe bit-error rate (BER) from the relationship among SNR,error vector magnitude (EVM) and BER for M-QAM pub-lished in [15]. The SNR is measured at varying steps over a50cm distance between the transmitter and receiver.
(a) (b)
(c) (d)
Fig. 10: Captured frames with embedded PSK modulated beaconsusing a high-speed receiver in (a) and (b) and a low-speed receiverin (c) and (d).
(a) (b)
(c) (d)
Fig. 11: Captured frames with embedded BPM modulated beaconsusing a high-speed receiver in (a) and (b) and a low-speed receiverin (c) and (d).
According to the limitation on LED dynamic range, themaximum obtained SNR of OFDM samples conditioned toform the PWM-like frame is 26dB, as shown in Fig. 8. As ex-pected, the obtained SNR values result in a reasonable BERperformance for different modulation orders. For instance,
a BER of 7 × 10−5 is obtained for a 64-QAM modulatedwaveform.
In order to evaluate the low capacity link, the low-speedreceiver is used to capture the waveform and compare itwith the high-speed one. In Fig. 9, the embedded beaconingfunctionality of the designed waveform is shown. The sametransmitted waveform is captured by the high and low-speedreceivers as shown in the figure. In Fig. 9a, the PWM-likeencoded beacon, OFDM samples and fixed PWM-like dutycycles for dimming control are illustrated, unlike Fig. 9b,where all PWM-like and OFDM harmonics are filtered outand only the beacons are captured. In this figure a screen-shot of the received waveform by the low-speed receiver,with 20kHz embedded beacon, is shown after concatenationof multiple beacon signals. The demodulation capability oflow capacity bits using a low-speed receiver is clear fromFigs. 10 and 11. In Fig. 10, an example of BPSK modulatedwaveform is illustrated. To the left a frame with 0o phaseshift PWM encoded beacon, i.e. “0” bit, is captured by ahigh-speed receiver while to the right a 180o phase shiftis received. In Figs. 10c and 10d, the low-speed receivereasily captures the beacon’s phases. A BPSK demodulatorcan be easily applied in MATLAB to capture the bit stream.Similarly, BPM modulated beacons are captured within theirdesignated time slots in Fig. 11. On the left, the beaconis placed within the first time slot to represent a “000” bitpattern while the beacon representing “011” is on the rightside. In both cases, the low-speed receiver successfully filtersout the PWM-like and OFDM harmonics without corruptingthe beacon’s phase or position within the frame.
V. CONCLUSION
In this paper, the design framework of a novel VLC-basedmulti-tier waveform is presented. This waveform serves as auniversal waveform with multi-service capabilities. It offershigh and low capacity streams that can be captured bydifferent types of receivers. In addition, it offers embeddedbeacons for positioning services,as well as, dimming control.Experimental evaluation is performed to demonstrate andvalidate such simultaneous services. Based on the experimen-tal results, the designed waveform can offer dimming controlover 60% of the LED dynamic range while maintaining areliable communication link of 64-QAM at BER of 7×10−5.Extended analysis of spectral characteristics of the waveformis considered for future work.
ACKNOWLEDGMENT
This work was supported primarily by the Engineering Re-search Centers Program of the National Science Foundationunder NSF Cooperative Agreement No. EEC-0812056.
REFERENCES
[1] Poushter, J. (2016). Smartphone ownership and internetusage continues to climb in emerging economies: Butadvanced economies still have higher rates of technology use.http://www.pewglobal.org/2016/02/22/ smartphone-ownership-and-internet-usage-continues-to-climb-in-emergingeconomies/. Accessed:2018-05-05.
[2] C. X. Wang, F. Haider, X. Gao, X. H. You, Y. Yang, D. Yuan, H. M.Aggoune, H. Haas, S. Fletcher, and E. Hepsaydir. Cellular architectureand key technologies for 5G wireless communication networks. IEEECommunications Magazine, 52(2):122–130, February 2014.
[3] M. Ayyash, H. Elgala, A. Khreishah, V. Jungnickel, T. Little, S. Shao,M. Rahaim, D. Schulz, J. Hilt, and R. Freund. Coexistence of WiFiand LiFi toward 5G: concepts, opportunities, and challenges. IEEECommunications Magazine, 54(2):64–71, 2016.
[4] C. Chen, W. Wang, J. Wu, H. Chen, K. Liang, L. Wei, Y. Hsu,C. Hsu, C. Chow, C. Yeh, et al. Visible light communicationsfor the implementation of Internet-of-Things. Optical Engineering,55(6):060501–060501, 2016.
[5] M. S. A. Mossaad, S. Hranilovic, and L. Lampe. Visible lightcommunications using OFDM and multiple LEDs. IEEE Transactionson Communications, 63(11):4304–4313, Nov 2015.
[6] A. F. Hussein and H. Elgala. Lightweight multi-carrier modulationfor IoT. Proc. SPIE 10559, Broadband Access CommunicationTechnologies XII, 10559W:10559 – 10559 – 10, 2018.
[7] Hao Dong, Hongming Zhang, Kai Lang, Bingyan Yu, and Minyu Yao.OFDM visible light communication transmitter based on LED array.Chin. Opt. Lett., 12(5):052301, May 2014.
[8] F. Yang, J. Gao, and S. Liu. Novel visible light communicationapproach based on hybrid OOK and ACO-OFDM. IEEE PhotonicsTechnology Letters, 28(14):1585–1588, July 2016.
[9] Tianxing Li, Chuankai An, Zhao Tian, Andrew T Campbell, andXia Zhou. Human sensing using visible light communication. InProceedings of the 21st Annual International Conference on MobileComputing and Networking, pages 331–344. ACM, 2015.
[10] Hany Elgala and Thomas DC Little. Reverse polarity optical-OFDM(RPO-OFDM): dimming compatible OFDM for Gigabit VLC links.Optics express, 21(20):24288–24299, 2013.
[11] A. F. Hussein, H. Elgala, and T. D. C. Little. Visible light com-munications: Toward multi-service waveforms. 15th IEEE AnnualConsumer Communications Networking Conference (CCNC), pages1–6, Jan 2018.
[12] R. Mesleh, H. Elgala, and H. Haas. On the performance of differentOFDM based optical wireless communication systems. IEEE/OSAJournal of Optical Communications and Networking, 3(8):620–628,August 2011.
[13] R. Mesleh, H. Elgala, and H. Haas. Performance analysis of indoorOFDM optical wireless communication systems. In IEEE WirelessCommunications and Networking Conference (WCNC), 2012.
[14] A. F. Hussein, H. Elgala, B. Fahs, and M. M. Hella. Experimentalinvestigation of DCO-OFDM adaptive loading using Si PN-basedreceiver. In 26th Wireless and Optical Communication Conference(WOCC). IEEE, April 2017.
[15] R. Chouitem. EVM based AMC for an OFDM system. 2010 WirelessTelecommunications Symposium (WTS), pages 1–5, April 2010.
